Yes, it may be a bug in ODL, but I also think that there is something wrong to 
use such a long relative path.  In that case, the relative path is actually 
longer than the absolute path!  That said, there is no guidance in rfc6087bis 
currently, so we may need to let it go this time.

K. // shepherd

===== original message =====

This may be. bug in OpenDaylight that is being tickled.  Ranga is
chasing it a bit.

On 09.02.18 10:20, Jan Lindblad wrote:
> Eliot,
>> In order to compile the published YANG model with OpenDaylight Yangtools I 
>> had to make the following change ( diff published file vs. changed file is 
>> below ):
>> 583c583
>> <                 path "../../../../../../acl/name";
>> ---
>>>                 path "/access-lists/acl/name";
>> 597c597
>> <                   path "../../../../../../../acl/aces/ace/name";
>> ---
>>>                   path "/access-lists/acl/aces/ace/name";
>> I am not sure (don't have enough YANG experience) to know if the error is 
>> with Yangtools or with the published YANG model but I thought I'd send this 
>> information to the list just in case.
>> Thank you for your attention.
> Both the old and the new path look valid to me. Even if you can always 
> replace a relative path with an absolute from a YANG validity perspective, 
> changing from relative to absolute paths often *changes the semantics*, so 
> that is not generally safe. In this case, however, they do amount to the same 
> thing (since they both end up going all the way up to the top level 
> container).
> /jan

netmod mailing list

Reply via email to